1. Trang chủ
  2. » Luận Văn - Báo Cáo

Báo cáo khoa hoc:" Genetic parameters of beef traits of Limousin and Charolais progeny-tested AI sires" pdf

25 325 0
Tài liệu đã được kiểm tra trùng lặp

Đang tải... (xem toàn văn)

Tài liệu hạn chế xem trước, để xem đầy đủ mời bạn chọn Tải xuống

THÔNG TIN TÀI LIỆU

Thông tin cơ bản

Tiêu đề Genetic Parameters of Beef Traits of Limousin and Charolais Progeny-Tested AI Sires
Tác giả Marie-Noëlle Fouilloux, Gilles Renand, François Mônissier, Jacques Gaillard
Trường học Institut National de la Recherche Agronomique
Chuyên ngành Agricultural Sciences
Thể loại bài báo
Năm xuất bản 1999
Thành phố Jouy-en-Josas
Định dạng
Số trang 25
Dung lượng 1,44 MB

Các công cụ chuyển đổi và chỉnh sửa cho tài liệu này

Nội dung

At the end of the test, the best bulls to be progeny tested were selected according to an index combining three or four traits recorded in these central stations.. Effect of step 2 selec

Trang 1

Original article

of Limousin and Charolais

Marie-Noëlle Fouilloux a Gilles Renand a Jacques Gaillard

François Ménissiera

Station de génétique quantitative et appliquée, Institut national

de la recherche agronomique, 78352 Jouy-en-Josas, France

Institut de l’élevage, Station de génétique quantitative et appliquée, 78352

Jouy-en-Josas, France

(Received 10 November 1998; accepted 3 September 1999)

Abstract - Sire selection efficiency depends on the knowledge of accurate genetic

parameters In France, artificial insemination (AI) sires are selected according to their

own performances and those of their progeny, which are both recorded in test stations.

Genetic parameters among progeny traits were estimated using multi-trait REML

(restricted estimation of maximum likelihood) analyses in Charolais and Limousinbreeds The expected decrease in genetic variability algebraically calculated among

progeny traits due to the selection of sires was not observed This selection was not a

strict truncation Heritabilities of traits measured on progeny are moderate for growth traits, morphology and live fatness scores (from 0.14 to 0.38) and slightly higher for

dressing percentage and carcass fatness score (0.50 and 0.44, respectively) Geneticcorrelations among progeny traits depended on traits, selection programme and breed.Carcass weight and morphology were highly genetically linked to corresponding livetraits (live weight and conformation, respectively) They can, therefore, be easily improved through indirect selection in contrast to carcass fatness which has only a

small genetic correlation with live traits © Inra/Elsevier, Paris

genetic parameters / live and carcass traits / Charolais and Limousin breeds /

selection

Résumé - Paramètres génétiques des aptitudes bouchères des taureaux nation artificielle Limousins et Charolais contrơlés sur descendance L’efficacité

d’insémi-de la sélection d’insémi-des reproducteurs dépend de l’exactitude des paramètres génétiques

utilisés En France, les taureaux d’insémination artificielle sont sélectionnés à partir

de leurs performances propres et celles de leurs descendants mesurées en station de

*

Correspondence and reprints

E-mail: Fouilloux@dga.jouy.inra.fr

Trang 2

contrôle Les paramètres génétiques des performances

en race Charolaise et Limousine à l’aide d’un REML (Estimation du Maximum deVraisemblance Restreint) - multicaractère La réduction calculée algébriquement de lavariabilité génétique des performances des descendants due à la sélection des pères, n’a pas été observée Cette sélection n’a pas été faite par troncature stricte L’héritabilitédes caractères de croissance, de morphologie et d’état d’engraissement est modérée

(comprise entre 0,14 et 0,38) Celle du rendement de carcasse et de la note de grasinterne est plus élevée (0,50 et 0,44, respectivement) Les corrélations génétiques dépendent, notamment, des caractères analysés, du programme de sélection et de la

race Le poids et la conformation des carcasses sont fortement corrélés génétiquement

à des caractères mesurables sur l’animal vivant Ils sont donc aisément améliorablespar sélection indirecte contrairement à l’état d’engraissement des carcasses qui n’apparaît que peu lié génétiquement aux caractères contrôlés en vif @ Inra/Elsevier,Paris

paramètres génétiques / caractères en vif et d’abattage / races Charolaise et Limousine / sélection

1 INTRODUCTION

In France, beef traits of artificial insemination (AI) bulls are improved by

a three-step sequential selection The first step is based on pedigree and formances at weaning The second step is based on post-weaning performances

per-of bulls recorded in central test stations The last step is based on the formances of a sample of the male progeny of these bulls fattened in progeny

per-test stations Breeding values of these sires for beef production are currentlyestimated using the latter two data sets [1].

Since the beginning of the 1980s, heritabilities of beef traits currently

used in genetic evaluation programmes in France have been based on theestimates given by Renand and Gaillard [29], Renand [25, 26] and Renand

et al [30] in different beef breeds, using the Henderson method 3 without a

relationship matrix among sires Since the accuracy of genetic evaluations andconsequently the efficiency of selection partly depend on the use of correct sound

parameters (heritabilities and genetic correlations), these estimates need to bereconsidered for two reasons: 1) more recent information is available in theseselection programmes; 2) variance component estimations can be obtained with

more suitable methods, such as restricted estimation of maximum likelihood

(REML), known to be the method of choice for most situations in animalbreeding Sire selection based on their own performance prior to their progenytesting was expected to modify the subsequent genetic variability [4, 8! Then,

an unbiased estimation of genetic parameters requires that the data used forselection decisions (performance and pedigree up to the base population) beincluded in the analysis (35! Journaux [13] estimated genetic parameters of a

trait observed for progeny and a trait observed for the sires using a bivariateREML approach Nowadays, such multivariate REML estimates could allow,

to a certain extent, the estimation of variance components taking into account

the information used for selection

The objective of this paper was to estimate the genetic parameters to beused for progeny testing after checking whether the previous selection of siresbased their performance should be taken into account.

Trang 3

MATERIAL AND

2.1 Design of testing procedures in the French AI programmes

In each of the specialised beef breeds in France, two types of programmesexist for selecting AI bulls depending on whether they are predominantly usedfor terminal crossbreeding or for pure-breeding.

Each year, new potential AI bulls were bought by AI co-operatives atweaning in nucleus herds and gathered in central test stations (50-70 per year

on average) The actual information used by AI co-operatives for selecting thesecalves was not known Two or three groups of contemporary calves (born within

a 6-week period) were then tested for a fixed period length up to approximately

16 months of age At the end of the test, the best bulls to be progeny tested

were selected according to an index combining three or four traits recorded

in these central stations These performances were final weight, feed efficiencyand muscling score for selecting terminal crossbreeding AI bulls Skeletal frame

score was added when AI bulls were used for pure-breeding [1] Semen quality

of selected bulls was assessed before progeny testing This selection step was

not a strict truncation (figure 1) because some sires with high indexes were

eliminated either for bad semen quality or other defects

Bulls selected (on average 8-13 per year) were randomly mated to about

100 adult cows in commercial herds Three reference bulls were simultaneouslyused Approximately 20-30 male calves per tested bull and per reference sires

at 15-20 days (crossbred) or 6-7 months (pure-bred) of age were bought and

set in the test stations Crossbred calves were raised in a nursery until the

beginning of the performance test (5-6 months) The performance test of thepure-bred calves started after 1 month of adaptation At the beginning ofthe performance tests, young bulls were gathered in age-contemporary groups(variation of 1 month maximum) During the test period, male calves were

intensively fattened with corn silage distributed ad libitum and supplementedwith protein feed They were slaughtered under uniform conditions at a fixed

age or fixed weight depending on the selection programme Carcass traits

were recorded In each progeny test station, batches for different years were

genetically connected through three national reference sires !1!.

young bulls (Normand and Friesian dams) slaughtered at a fixed weight of

600-650 kg depending on the year.

A total of 131 Limousin and 145 Charolais sires was progeny tested on 4 532and 3 519 young bulls, over 11-12 consecutive years, respectively Most of these

sires were previously tested in central test stations

Trang 4

2.3 Performances recorded in progeny test station

Owing to the strict procedures and the restricted number of animals inthe station, many performances concerning growth and conformation could be

Trang 5

accurately recorded before or after slaughter The beef traits analysed in thisstudy were:

slaughter yield: dressing percentage (DP) defined as the ratio of hot carcass

weight to final live weight;

-

morphology scores: live muscling (LM), carcass muscling (CM) and liveskeletal frame (LS) scores;

- fatness scores: live fatness (LF) and carcass fatness (CF) scores As

carcasses were systematically trimmed at slaughter, CF was scored for the

amount of pelvic, kidney and internal fats

Scores were given by a very limited number of experienced technicians in

each station at the very end of the test period (LM, LS, LF) and at slaughter

(CM, CF).

2.4 Effect of selection of sires on the genetic variability of progeny

traits

2.4.1 Effect of step 2 selection

In order to study the impact of the selection of sires (step 2) on the genetic variability of progeny traits three different estimates of genetic parameters were

compared This selection was based on the sire own performances measured

in the central test station In the Charolais programme, a set of four traits

measured on progeny was studied: two live traits (live weight (LW) and livemuscling score (LM)) and two slaughter performances (dressing percentage

(DP) and carcass fatness score (CF)).

- The first estimates (h and r ) were obtained on these four progeny traitsanalysed simultaneously with the three performance traits of sires used forselecting bulls on their own performances in the test station (final weight,feed efficiency and live muscling score) The progeny trait (co)variances were

described with a sire model while the sire performance (co)variances were

described with an animal model Since all the data presumably used forselecting the sires were included in the analysis, these estimates were considered

to be free from the influence of selection in step 2

- The second estimates (ha and r ) were obtained on the four progenytraits only, described with a sire model These apparent genetic parameters

might have been biased by selection

- The third estimates (hfl and rgg) were algebraically derived from the first

ones taking into account the reduction in variance of traits among selectedsires Selection at the end of the performance tests in the central test station(step 2) was assumed to be only made on a selection index combining finalweight (P ), live muscling score (P ) and feed efficiency (P ) A posteriorindex (1) [20] was obtained from the observed selection differentials of each

trait (Pi - Pi), where Pi and Pi, were the means of sires for the ith trait

Trang 6

before and after selection:

Using this index with a threshold selection would have led to the observedselection differential for each trait

In the Charolais programme, 118 out of the 145 progeny-tested sires were

selected among 519 bulls tested in the central station The observed selectiondifferential was about 7.00 on that posterior index (I) Because the observed

variance before selection ( ) was 57.3, selection intensity was equal to 0.93.The variance observed among selected bulls (o, 2!, was 24.3 (43 % of a)) andthe relative reduction of variance, (3 = (o, 2s -ol 2)/0,2 was -0.58

Such an investigation was carried out in the Limousin programme, where

112 sires were progeny tested out of 470 bulls tested in the central station

Similarly to the Charolais analysis, three sets of genetic parameters amongprogeny live weight, live muscling score, dressing percentage and carcass fatness

score were estimated according to different models considering or not theselection of bulls in the performance test station Limousin bulls were selectedaccording to their final weight (FW), feed efficiency (FE), live muscling score

(LM) and skeletal frame score (LS) A posterior index was calculated combiningthe FW, FE, LM and LS The observed selection differential was about 6.34for that posterior index (1) with a selection intensity of 0.96 The observedvariances of the posterior index (I) before and after selection were equal to

a) = 43.2 and U2 &dquo; = 25.5, respectively (0,2,/Ol = 59 %) The relative reduction

of the index variance ((3) was equal to -0.41

Knowing the weights (b ) of traits (i) in the selection index (I =

L b

i

the relative reduction of index variance (!3) and the correct genetic parameters (h and rg), the genetic parameters in this sample of selected sires that were

expected to be observed (h e 2 Iek h2 and re9!!) for progeny traits (j and k) can

be calculated algebraically The formulae initially given by Robertson [32] forsingle trait selection were extended to a selection on a selection index [23] (see Appendix):

Trang 7

where Q9! was the genetic standard deviation of trait i in the sire selectionindex.

2.4.2 Effect of step 1 selection

As bulls were previously selected according to some information at weaningbefore being performance tested in the station, the genetic variability of traits

measured on progeny might eventually have been affected by that step 1selection

In order to estimate the impact of selection at weaning on the progenygenetic parameters, weaning performances of all contemporary males raised inthe same herds should be considered Performances at weaning of male calvesfrom the selected bull’s contemporary-herd group were extracted from a data

set used in a French beef bull evaluation programme on performances recorded

in farms [1] In the Charolais breed, weaning performances of 15143 young

bulls were available (419 tested in the central station) In the Limousin breed, weaning performances of 14 909 young bulls were available (407 tested in thecentral station) Such an amount of information prevents one from integrating weaning traits together with progeny traits in a multiple trait analysis forestimating genetic parameters free of that selection effect It was only possible

to use the algebraic formulae (1 and 2) for predicting what should have beenthe impact of that selection The use of these formulae required, however, that

true j3 and genetic parameters be known

As the actual criteria used to choose tested bulls were unknown, an intensity

of selection at weaning was calculated postulating that AI co-operatives didselect the male calves according to weight (WW), muscularity (WM) and, in

the Limousin programme, skeletal frame (WS) at weaning.

’Superiority’ of each selected male was calculated as the standardised ference between its performances (WW, WM or WS) and the average of malecalves from its contemporary-herd group:

dif-where Sh!i was the ’superiority’ for trait i (i = WW, WM or WS) of selectedcalf j raised in the contemporary-herd group h; Ph!i was the performance ofthis calf j for trait i; P was the mean of the male contemporary-herd group

h; and Qhi was the standard deviation in this group.

A posterior selection index (I) was calculated, combining these ’superiorities’for WW, WM and, in the Limousin programme, WS

In the Charolais programme, the observed selection differential was about35.5 for that posterior index (1) and the observed variance ( ) before selection

was 400.0 Hence, the observed selection intensity was equal to 1.78 Thevariance of the posterior index (I) after selection was a 2!, = 186.0 (a =

47 %) The relative reduction of the index variance (0) was therefore equal to

- 0.54

In the Limousin programme, the observed selection differential was about30.3 for that posterior index (I) and the observed variance (!1) before selection400.0 Hence, the observed selection intensity was equal to 1.52 The

Trang 8

variance of the posterior index (I) after selection was ay 179.3 ( a2 1 =

45 %) The relative reduction of the index variance (0) was equal to -0.55.Since the true genetic correlations (ri! ) between weaning traits on farms andtraits recorded in the progeny test were not known, those estimated between

post-weaning traits of sires (LW for WW, LM for WM and LS for WS) andtraits recorded in the progeny test measured in the stations (tables II and III)

have been considered as the soundest correlations Heritabilities (h! ) of progenyperformances estimated jointly to the sire’s own performance (see section 2.4.1)

were considered as the most reliable

2.5 Models of analysis and methods

2.5.1 Models

In both breeds, the models of analysis of progeny traits included fixed

envi-ronmental effects and random sire effect(s) There were no genetic relationships

among dams and between dams and sires Genetic relationships among sires

took into account up to two generations of ancestors.

The following models were used, subsequently to an analysis of variance thattested the significance of the fixed effects (General Linear Model, SAS).

In both breeds, the main fixed effects were calving parity of the dams

(calv: 2, 3, 4, 5 and over), region of origin (orig) and age-contemporary group

(cont) of the young bulls Age-contemporary groups corresponded to age-test

groups in the station Other fixed effects included: in the Limousin model, a

management system up to weaning (manag: indoor or outdoor weaned calves);

in the Charolais model: the breed of the dam (breed: Holstein-Friesian or

Normand) and health status in the nursery (pulm and diges: occurrence or

Trang 9

absence of pulmonary digestive troubles) In both breeds, average daily gain

was regressed on initial age (¡3Cov) Muscling (LM, CM), skeletal (LS) andfatness (LF, CF) scores were regressed on final age in the Limousin breed and

on final weight in the Charolais breed (¡3Cov).

Ch.: yij =

cont+ calv+ orig+ breedij + pulm+ diges2! + ¡3C ij + si

Lim.: y2! =

cont + calv+ orig + manage !- ¡3C ij !- sz

y2! was the performance of the jth male progeny of the ith sire

In the study of the sire step 2 selection effect, sire performances were analysed

in an animal model (a) with an age-contemporary group (cont) In the Charolaisbreed, pre-test environment (pre-test) fixed effects were added A regression on

final age (¡3Age) was performed for both breeds:

Trang 10

Ch.: y cont+ pre-test + ,!Agei + a

Lim.: y = cont + (3Age+ ai

Y was the performance of the ith sire

Trang 11

likelihood (REML) method in multi-trait (nine traits) analysis for each breed

(VCE3.2 Package, Groeneveld !11!) These components allowed the elaboration

of phenotypic correlations (r ) and genetic parameters, heritabilities andgenetic correlations (ha and r

In the study of the sire selection effect, the variance and covariance

compo-nents among and between sire performances and progeny traits were estimated

in a multi-trait analysis (four progeny traits (LW, LM, DP, CF) with or

with-out three or four sire traits (FW, LM, FE and, in the Limousin analysis, LS)

by the REML method using VCE4.2 of Groeneveld !11!.

3 RESULTS AND DISCUSSION

3.1 Means and phenotypic variability (table I)

In the Limousin breed, the variability of initial weight was especially high

(SD = 54 kg and CV = 15 %) The variability of the corresponding traitamong the Charolais crossbred calves was moderate (SD = 24 kg and CV =

11 %) Another analysis using Charolais pure-bred young bulls brought into theprogeny test station at weaning, gave high initial weight variability (SD = 61 kgand CV = 15 %) Raising animals in a common environment might contribute

to reducing the differences between animals due to pre-test conditions In both

progeny test stations the variability of live weight around 15 months (LW: CV

= 7 and 9 % in the Charolais and Limousin programmes, respectively) was

lower than the initial weight variability (IW: CV = 11 and 15 %, respectively).

Fattening progeny in a common environment reduced phenotypic variability

among young bulls.

Variability of dressing percentage was low, similar to most of the results in

the literature

Variability of morphology and fatness scores was relatively high with cients of variation between 9 and 18 %.

coeffi-The slaughter point criteria was expected to be fixed in both programmes:

age in the Limousin and weight in the Charolais programme This was only partially obtained, especially in the Charolais breed There were only 4 days forwithin-year standard deviation (CV < 1 %) of slaughter age in the Limousin

programme versus 17 kg (CV = 3 %) for slaughter weight in the Charolaisprogramme A fixed age end point is clearly easier to organise than a fixedweight.

3.2 Effect of selection of sires on the genetic variability of progenytraits (tables II and III)

3.2.1 Effect of step 2 selection

In both programmes, the apparent genetic parameters (ha and r g)

esti-mated without considering the effect of the previous selection of sires were

close to the sound ones (h and r ) estimated jointly with sire performancedata (Charolais: table II; Limousine: table III) Apparent heritabilities were

only slightly lower (by 0.02-0.03) than correct ones, and genetic correlations

Trang 12

within 0.03 of sound These differences depend the relative change

of variance of the traits under selection (0), the true heritability of these traitsand the magnitude and the sign of the true genetic correlations between these

traits and progeny traits In the present study, the relative reductions of theselection criteria variance, (3 (o, 28 _ol 2)/0,2, used for computing the expected genetic parameters were negative (Charolais: - 0.58; Limousine: - 0.41) sinceunilateral selection reduced the index variance Fimland [8] showed that theexpected effect of selection for performance is generally minor for heritabilities

For example, with close to - 0.55, and heritability close to 0.30 and 0.40,

he predicted that the remaining genetic variance still represents about 95 %

of the initial genetic variance On the contrary, a substantial modification ofthe genetic correlation (r9!k) between two progeny traits may be expected if

at least one of these traits has a close genetic correlation with the selectioncriteria Usually, with a negative ,(3 the change in r9!k is negative when rg

and rg have the same sign and the change is positive when rgjj and r have

When the apparent genetic parameters were compared to the expected ones

after algebraic correction, no difference was found for live muscling scores,

dressing percentages and carcass fatness scores in either programme Differencesfor heritabilities were smaller than 0.02 and differences for genetic correlations

were smaller than 0.05 However, the differences for live weight (LW) were

larger For example, the corrected heritability of LW was 0.07 lower than the

apparent one in both programmes Therefore, the formulae used for predictingthe effect of selection were not adapted to the real selection procedure fordifferent possible reasons The genetic correlations between sire and progenytraits used in the algebraic formulae might not have been exactly the true

ones as they were estimated with low accuracy (average standard errors equal

to 0.10) The step 1 selection could not have been taken into account sinceadding the corresponding amount of information (the whole contemporary

group in the herd of origin) made a joint analysis unfeasible as it wouldhave exceeded the capacity of our data processing This step 1 selectionmight have influenced the estimated genetic parameters among progeny traits

and between sire and progeny traits The selection procedure (step 2) was

certainly not a threshold selection on a selection index combining only three

sire performances (LW, LM, FE) because some sires with high indexes were

eliminated either for some defects (semen quality, foot and leg soundness, etc.)

or for breed standard criteria Consequently, some bulls with low indexes were

kept (figure 1) Contrarily to the estimation of apparent genetic parameters

that used information available on progeny of 145 sires, the expected genetic

parameters were calculated using the progeny of only 118 sires that were

previously tested in the same central station The remaining 27 sires (! 19 %)

had various origins In the Limousin programme, only 112 sires among the

131 that were progeny tested were previously selected in a performance testingstation Such a loss of information might have influenced the expected geneticcorrelations

Ngày đăng: 09/08/2014, 18:21

TỪ KHÓA LIÊN QUAN

🧩 Sản phẩm bạn có thể quan tâm